2019年国际光伏路线图（英文）

AIP Conference Proceedings 2147, 140005 (2019); https://doi.org/10.1063/1.5123892 2147, 140005 © 2019 Author(s). Impact of the thermal budget of the emitter formation on the pFF of PERC+ solar cells Cite as: AIP Conference Proceedings 2147, 140005 (2019); https://doi.org/10.1063/1.5123892 Published Online: 27 August 2019 Philip Jäger, Ulrike Baumann, and Thorsten Dullweber ARTICLES YOU MAY BE INTERESTED IN Study on the interfacial oxide in passivating contacts AIP Conference Proceedings 2147, 040016 (2019); https://doi.org/10.1063/1.5123843 Fine line Al printing on narrow point contact opening for front side metallization AIP Conference Proceedings 2147, 040019 (2019); https://doi.org/10.1063/1.5123846 PERC+ solar cells with screen-printed dashed Ag front contacts AIP Conference Proceedings 2147, 060001 (2019); https://doi.org/10.1063/1.5123861 Impact of the Thermal Budget of the Emitter Formation on the pFF of PERC+ Solar Cells Philip Jäger 1, a) , Ulrike Baumann 1) , Thorsten Dullweber 1) 1 Institute for Solar Energy Research Hamelin (ISFH), Am Ohrberg 1, Emmerthal, Germany a) Corresponding author: jaeger@isfh.de Abstract. We develop processes for advanced phosphorus doping profiles in order to reduce the emitter saturation current density Jo,e of industrial bifacial PERC+ solar cells. With an in-situ oxidation, which takes place in the POCl3 furnace in between the deposition and the drive-in step, the surface concentration was lowered from 3 × 10 20 cm -3 to 1.7 × 10 20 cm -3 . With an additional ex-situ oxidation, which takes place after the phosphorus silicate glass is removed, the phosphorus surface concentration was further reduced to 3 × 10 19 cm - 3 . The decreased phosphorus surface concentration drastically reduces Jo,e from 106 fA/cm 2 down to 22 fA/cm 2 . The reduced Jo,e increases the implied open circuit voltage up to 712 mV of unmetallized PERC+ test structures and the Voc of PERC+ solar cells up to 678 mV and efficiencies up to 21.8%. However, our solar cell analysis reveals for the first time, that with increasing thermal budget of the applied POCl3 and oxidation recipes the pseudo fill factor (pFF) decreases by up to 1.5%. This corresponds to an efficiency loss of approximately 0.5%abs. We analyse the pFF loss based on different lifetime test structures representing the emitter or the bulk of the PERC+ solar cell. From the lifetime measurements we calculate I-V curves representing the implied fill factor (iFF) of the different parts of the PERC+ solar cell as well as a combined one for the whole cell, which compares well to the measured pFF. The iFF values clearly show that the pFF is mainly limited by wafer bulk material. However, also the iFF values of the emitter slightly decrease with increasing thermal budget. INTRODUCTION The International Technology Roadmap for Photovoltaics (ITRPV) just published a current market share for PERC solar cells of 50% and expecting a market share of 75% in 2026, which may include PERC cells with passivating contacts [1]. The conversion efficiency of industrial PERC and bifacial PERC+ solar cells [2] is mainly limited by charge carrier recombination in the phosphorus-diffused emitter [3]. In order to reduce the emitter saturation current density J o,e of PERC solar cells, in recent years new POCl 3 recipes have been developed applying additional oxidation steps. One approach is a so-called in-situ oxidation, where the oxygen flow continues after the phosphorus silicate glass (PSG) deposition, thereby growing an interfacial oxide between the PSG and the silicon [4][5]. The interfacial oxide limits the diffusion of phosphorus from the PSG into the silicon wafer [5] and thus lowers the phosphorus surface concentration and over all phosphorus amount in the wafer. The in-situ oxidation enables J 0,e values down to 22 fA/cm 2 [6]. The interfacial oxide is etched back during the PSG etch and therefore does not serve as a passivation layer. Another option is an ex-situ oxidation, where a thermal oxidation around 800 to 900 °C is carried out after the PSG was etched of [4]. The ex-situ oxidation drives the phosphorus deeper into the silicon thereby substantially lowering the phosphorus surface concentration. It also oxidises a few nanometres of the silicon surface, which provides a high-quality SiO 2 surface passivation. In this contribution, we develop and optimize different in-situ and ex-situ oxidized POCl 3 emitter recipes with the original motivation to reduce the emitter recombination in industrial PERC+ cells as described above. However, the PERC+ solar cells reveal that the different emitter recipes not only affect the open-circuit voltage but also strongly influence the pseudo fill factor pFF. We investigate this novel effect based on different lifetime test structures, which represent the carrier recombination in different parts of the PERC+ solar cells. SiliconPV 2019, the 9th International Conference on Crystalline Silicon Photovoltaics AIP Conf. Proc. 2147, 140005-1–140005-6; https://doi.org/10.1063/1.5123892 Published by AIP Publishing. 978-0-7354-1892-9/$30.00 140005-1 RECIPES AND SAMPLE PREPARATION We process industrial bifacial PERC+ cells as shown in Fig. 1 a) and b). We use 1.5 Ohm·cm boron-doped CZ wafers with front side texture, n+ phosphorus emitter, AlO y /SiN x rear side passivation, laser contact opening and screen-printed metal contacts. Overall we compare 6 groups: two groups without oxidation steps, one group with only an in-situ oxidation and three groups, which had the in-situ oxidation as well as an ex-situ oxidation. An overview of the emitter formation is shown in Table 1. For the first group ‘Without oxidation 1’, we use a diffusion process without an oxidation step and the temperature was kept constant at 829 °C for deposition and drive-in. The wafers are textured and diffused on both sides followed by a rear side polish to remove texture and emitter. In contrast, the group ‘Without oxidation 2’ was processed with a rear side SiN x protection layer, so that only the front side is textured and diffused. After PSG and dielectric etch, the process was carried out like that one of the previous group but without busbars. The group ‘In-situ oxidation’ is processed like ‘Without oxidation 1’ but receive an additional 5 minutes oxidation before and another 30 minutes after the drive-in under nitrogen flow. The front sides of these three groups are passivated by a PECVD silicon nitride (SiN x ) after the PSG removal. The group ‘Ex-situ oxidation 1’ receive the in-situ oxidation and, after the PSG was etched of, is additionally oxidized at 900 °C for 15 minutes and is left in the oven for 15 minutes under nitrogen atmosphere. The grown 9 nm thin thermal oxide are kept on the front side as surface passivation and a PECVD SiN x is deposited on top. The group ‘Ex-situ oxidation 2’ was processed identical to the previous group but the ex-situ oxidation temperature is set at 950 °C. The last group, ‘Ex-situ oxidation 3’, is processed the same way, but at an oxidation temperature of 850 °C and without busbars. For the four groups with oxidation steps we form a selective emitter by laser doping the PSG with a green laser. The laser doping increases the phosphorus concentration below the silver fingers enabling lower contact resistances. Process step w/o oxidation 1 w/o oxidation 2* In-situ oxidation Ex-situ oxidation 1 Ex-situ oxidation 2 Ex-situ oxidation 3* POCl3 deposition 30 min 30 min 30 min 30 min 30 min 30 min 829 °C 829 °C 829 °C 829 °C 829 °C 829 °C In-situ O2 flow - - 5 min 5 min 5 min 5 min N2 flow 70 min 70 min 70 min 70 min 70 min 70 min In-situ O2 flow - - 30 min 30 min 30 min 30 min Ex-situ O2 flow - - - 15 min 15 min 15 min 900 °C 950 °C 850 °C Ex-situ N2 flow - - - 15 min 900 °C 15 min 950 °C 15 min 850 °C FIGURE 1. Schematic illustration of the bifacial industrial PERC+ solar cells for a) Groups without an ex- situ oxidation with SiNx as front side passivation and b) for the groups with ex-situ oxidation and the resulting thin thermal oxide and SiNx as passivation TABLE 1. Process parameters for the split groups. Differences from the previous group are bold. The asterix (*) indicates, that these groups are processed without busbars 140005-2 We measure the cells with busbars on the IV measurement system LOANA from PV-Tools and the ones without busbars on a SpotLight system from Meyer Burger. To investigate the impact on the pFF two different lifetime test structures are used as shown in FIG. 2 a) and b). For the emitter a symmetrical test structure is processed, using 3.3 ohm·cm n-type wafers with texture, POCl 3 diffusion and to the cells corresponding front side passivation. For the three groups with ex-situ oxidation an additional thin thermal oxide is present, which is not shown in the picture. The test structures for the wafer bulk are also symmetrically processed and apply the same 1.5 ohm·cm p-type CZ wafers as for the PERC+ cells. These wafers receive the emitter formation processes according to TABLE 1. Afterwards, the phosphorus emitter is removed on both sides by KOH etch followed by an RCA clean. Both wafer surfaces are passivated with a stack of AlO y and SiN x . Finally, both test structures are fired at 800°C in a belt furnace. The lifetime of the test structures is measured using the Quasi Steady State Photo Conductance (QSSPC) method. We calculate the suns-iV OC curves using the formulas (4) and (6) from [7] for each of the test structures and interpolate the suns-iV OC curves to calculate a combined suns-iV OC curve from both of the test structures to get a combined iFF. These we compare with the pFF of the PERC+ solar cells. Regarding the test structure in Fig. 2 a) we assume that the n-type material causes little recombination and therefore the emitter recombination determines the measured carrier lifetime. Because we have the emitter on both sides of the test structure and want to include only one of them into our calculation, we weight this test structure with a factor of 0.5. For the lifetime test structure in Fig. 2 b), we assume that the AlO x /SiN y passivation performs very well and that hence the wafer bulk material dominates the carrier lifetime of this test structure. Because we want to include the whole bulk, we weight this test structure with a factor of 1. This assumptions result in calculating the combined suns-iV OC curves which is intended to represent the final PERC+ solar cell. In addition, we use unmetallized PERC+ cells, that have the same process as the completed cells but without laser contact opening and screen printing of the metal contacts (see Fig. 3 a) and b)). From these we measure the implied open circuit voltage values by QSSPC, revealing the V oc potential of the different emitters. For our final analysis, we process PERC+ solar cells and the lifetime test structures in Fig. 2 using five different Cz wafer materials applying the ‘In-situ oxidation’ recipe to further investigate the impact of the wafer bulk on the pFF. Three of the wafer materials are doped with boron and two are doped with gallium. The first of the boron-doped material is the one we use for the PERC+ cells with base resistivity of 1.5 ohm·cm. Another one has a base resistivity of 1.7 ohm·cm and the last one has 1.1 ohm·cm. The first gallium-doped material shows a base resistivity of 1.0 ohm·cm while the second one has 1.4 ohm·cm. FIGURE 2. Schemes of the lifetime test structures: a) representing the emitter and b) representing the bulk FIGURE 3. Schemes of the implied open voltage test structures. They are processed like the completed PERC+ cells but without laser contact opening and screen-printing of the metal contacts. a) For the groups without ex-situ oxidation and b) with a thin thermal oxide after ex-situ oxidation 140005-3 EXPERIMENTAL RESULTS The oxidation steps reduce the phosphorus surface concentrations from 3×10 20 cm -3 without oxidation to 1.7×10 20 cm - 3 with in-situ oxidation and 3×10 19 cm -3 for both ‘Ex-situ oxidation 1 and 2’. Following this trend, the emitter saturation current density J 0,e decreases from 106 fA/cm 2 for the POCl 3 diffusion without oxidation down to 28 fA/cm 2 for the ex-situ oxidation 1 and 2 as shown in Fig. 4. When further optimizing the ex-situ oxidation at a temperature of 850°C in the group ‘Ex-situ oxidation 3’, we obtain a J o,e value of 22 fA/cm 2 (not shown in the graph). Whereas in Ref. [6] a J o,e value of 22 fA/cm 2 was obtained with a sheet resistance R sheet of 150 Ohm/sq. now we obtain the same J o,e value at a lower R sheet of 133 Ohm/sq. which causes less resistive losses when applied to PERC+ solar cells. The iV OC values strongly increase from 664 mV for the group ‘Without oxidation 1’ to 712 mV for ‘ex-situ oxidation 1’ (see Fig. 5). As shown in Fig. 6 the conversion efficiency increases from 20.7% for the POCl 3 diffusion without oxidation to 21.3% for the ‘Ex-situ ox. 1’ emitter, which is particularly caused by a strong increase in V oc from 655 mV to 678 mV. The J sc values increase from 39.3 mA to 39.7 mA. At the same time, the FF decreases from 81% for the groups ‘Without oxidation’ to 79.2% for the group ‘Ex-situ oxidation 2’, which was exposed to the highest thermal budget. Whereas the series resistance of the PERC+ cells remain quite constant and account for only 0.5% abs. FF decrease, the main share of 1.5% abs. FF reduction is caused by a strong decrease of the pseudo fill factor pFF with increasing thermal budget (see Fig. 7). The decreased pFF corresponds to a penalty in efficiency of approximately 0.5% abs . If processed without busbars the J sc increases by 1.2 mA/cm 2 for the diffusion without oxidation and of 0.5 mA/cm 2 for ex-situ oxidized ones. Here we are able to demonstrate efficiencies of 21.9% for the process ‘Without oxidation 2’ and 21.8% for the process ‘Ex-situ oxidation 3’. By modelling the IV curves of the PERC+ solar cells with a two-diode model with diode ideality factors of 1 and 2, we find that the decreasing pFF is caused by an increasing saturation current density of the second diode J 02 with increasing thermal budget from 4 to 9 nA/cm 2 (not shown in a graph). We then investigated the origin of the pFF loss based on iFF values derived from the lifetime test structures in FIG. 2 a) and b) with the method described above. Figure 8 a) shows the pFF obtained from the IV measurement of the PERC+ cells compared to the iFF of the emitter (test structure of Fig. 2a)), the iFF of the bulk (test structure of Fig. 2b)) and the calculated combined iFF of emitter (weighing 0.5) and bulk (weighing 1). The combined iFF values reproduce the course of the pFF quite well. The iFF values around 81% of the wafer bulk are much lower than the iFF values of the emitter around 84% and hence limit the combined iFF value. However, the trend of decreased iFF with increasing thermal budget is most clearly observed in the iFF values of th